MOLECULAR PHYLOGENETICS AND EVOLUTION Molecular Phylogenetics and Evolution 30 (2004) 545–562 www.elsevier.com/locate/ympev

Phylogeny and historical biogeography of diving (Coleoptera) inferred from mitochondrial DNA sequences

Ignacio Ribera,a,* Anders N. Nilsson,b and Alfried P. Voglera,c

a Department of Entomology, The Natural History Museum, Cromwell Road, London SW7 5BD, UK b Department of Biology and Environmental Science, University of Umea, SE 90187, Umea, c Department of Biological Sciences, Imperial College London, Silwood Park Campus, Ascot SL5 7PY, UK

Received 7 November 2002; revised 27 May 2003

Abstract

The Agabinae, with more than 350 species, is one of the most diverse lineages of diving beetles (). Using the mito- chondrial genes 16S rRNA and cytochrome oxidase I we present a phylogenetic analysis based on 107 species drawn mostly from the four main Holarctic genera. Two of these genera ( and Ilybiosoma) are consistently recovered as monophyletic with strong support, is never recovered as monophyletic, and is found paraphyletic with respect to several of the species groups of Platambus. Basal relationships among the main lineages are poorly defined, although within each of them relationships are in general robust and very consistent across the parameter space, and in agreement with previous morphological analyses. In each of the two most diverse lineages (Ilybius and Agabus including part of Platambus) there is a basal split between Palearctic and Nearctic clades, estimated to have occurred in the late Eocene. The Palearctic clade in turn splits into a Western Palearctic clade and a clade containing mostly Eastern Palearctic species, and assumed to be ancestrally Eastern Palearctic but with numerous transitions to a Holarctic or Nearctic distribution. These results suggest an asymmetry in the colonization routes, as there are very few cases of transcontinental range expansions originating from the Nearctic or the Western Palearctic. According to standard clock estimates, we do not find any transcontinental shift during the Pliocene, but numerous speciation events within each of the continental or subcontinental regions. Ó 2003 Elsevier Inc. All rights reserved.

1. Introduction cludes more than 350 known species in 10 genera, which are mostly distributed in the Holarctic, with only a few Dytiscidae diving beetles, with ca. 3800 known spe- small genera and species groups occurring in South cies (Nilsson, 2001), comprise one of the major lineages America, sub-Saharan Africa, and the Oriental and of aquatic Coleoptera. They are highly modified for Australian regions (Nilsson, 2000, 2001). living in a diversity of continental waters both as larvae The Agabinae is a morphologically homogeneous and adults, showing a wide range of ecological strategies group exhibiting few characters useful for systematics mainly reflected in different swimming behaviors and (Nilsson, 2000). They were traditionally considered a their associated morphotypes (Balke, in press; Ribera tribe within the subfamily , and placed and Nilsson, 1995). Dytiscids have a world-wide distri- among the basal lineages of Dytiscidae (see, e.g., bution, with the highest species diversity in the tropical Guignot, 1933). Phylogenetic analyses based on 18S and subtropical regions (Balke, in press; Nilsson, 2001). rRNA sequences confirm this basal placement, although However, a few north temperate lineages are very di- the group does not form a monophyletic clade with verse, such as the subfamily Agabinae. The group in- Colymbetini (Ribera et al., 2002a). This is consistent with results obtained by Miller (2001) based on mor- phological characters, who accordingly raised Agabini * Corresponding author. Fax: +44-207-942-5229. to subfamily level. The Agabinae are characterized by E-mail address: [email protected] (I. Ribera). the presence of a linear group of short, stout setae near

1055-7903/$ - see front matter Ó 2003 Elsevier Inc. All rights reserved. doi:10.1016/S1055-7903(03)00224-0 546 I. Ribera et al. / Molecular Phylogenetics and Evolution 30 (2004) 545–562 the posterior ventroexternal angle of the metafemur Outgroups included 24 species of all major lineages of (Brinck, 1948; Nilsson, 2000), although some species Dytiscidae with the exception of Hydroporinae and have apparently secondarily lost this character (e.g., Laccophilinae, plus families Hygrobiidae and Amphi- Hydronebrius and some species of Platambus, Nilsson, zoidae (Appendix A). This extensive selection of out- 2000). Attempts to establish natural groups were made groups was considered necessary, as the phylogenetic only recently, but because of the homogeneous mor- position of Agabinae among the basal Dytiscidae lin- phology the delimitation of the Agabinae genera re- eages is still uncertain. All trees were rooted in Amphi- mains problematic (Nilsson, 2000). Two major groups zoidae, which clearly belongs outside Dytiscidae (Miller, of genera can be separated within the subfamily, ac- 2001; Ribera et al., 2002a,b). Rooting the tree in Am- cording to the presence of a ventral setal fringe on the phizoidae allowed the relationships among the remain- female metatibia and the shape of the clypeal fovea: the ing outgroups and the ingroup to vary freely, thus group (four genera with a Neotropical and testing the possibly monophyly of Agabinae and the Oriental distribution) and the Agabus group (six genera Agabus group of genera. with mostly Holarctic distribution). Within these gen- Specimens were collected in absolute ethanol, and era, species groups have been defined on the base of muscular tissue used for DNA isolation via a phenol– character combinations, with no robust synapomorphies chloroform extraction as described in Vogler et al. except for some subgenera of Agabus (Nilsson, 2000). (1993) or by extraction columns (Quiagen). Sequences of The Agabinae not only constitute a major element of 16S rRNA (16S) were amplified as a single fragment of the tree of Dytiscidae, they are also of interest to the study ca. 500 bp, using primers 16Sa (50 ATGTTTTTGTT of species richness and historical biogeography. Species AAACAGGCG) and 16Sb (50 CCGGTCTGAACTC differ greatly in the overall size of their geographic ranges, AGATCATGT). A single fragment of ca. 800 bp of COI with some species distributed very widely throughout (from the middle of region E3 to the 30 end, Lund et al., much of the Palearctic, Nearctic, or both. On the other 1996) was amplified using primers ‘‘Jerry’’ (50 CAACA end of the spectrum, some species are limited to narrowly TTTATTTTGATTTTTTGG) and ‘‘Pat’’ (50 TCCAAT defined biogeographic regions, in particular those species GCACTAATCTGCCATATTA) (Simon et al., 1994). It occurring in mountain regions in the southern Palearctic was not possible to obtain PCR products for the COI (e.g., Fery and Nilsson, 1993; Guignot, 1933). fragment of five species (A. affinis, I. bedeli, I. apicalis, In this paper we present mtDNA data for all major P. glabrellus,andA. nannup), and for three species lineages of Agabinae and a large number of the known (A. affinis, A. lapponicus,andP. pictipennis) portions of species, with an emphasis on the Holarctic fauna. We the sequence at the beginning or the end of the sequence also address the main geographical divisions between were missing. For the 16S fragment all specimens were the Palearctic and Nearctic regions, and provide an successfully amplified, except for a small number of base approximate temporal framework for the evolution and pairs at either end of the fragment in some species. All the main vicariance events affecting lineages on both new sequences generated in the study were deposited in sides of the northern Atlantic and Pacific oceans. GenBank under Accession Nos. AY138591–AY138768. Other sequences were obtained from Ribera et al. (2001a,b, 2002b) (Appendix A). 2. Material and methods The following PCR cycling conditions were used for DNA amplification: 10–20 at 95 °C, 3000 at 94 °C, 3000 at 2.1. Taxon sampling and DNA sequencing 47–50 °C (depending on the melting temperatures of the primer pair used), 10–20 at 72 °C (repeated for 35–40 The ingroup included a total of 118 individuals from cycles), and 100 at 72 °C. Amplification products were 107 species in five genera of Agabinae, with representa- purified using a GeneClean II kit (Bio 101). Automated tion of all major lineages (Table 1 and Appendix A). DNA sequencing reagents were supplied by Perkin Sampling was focussed on the larger genera of the Ag- Elmer Applied BioSystems (ABI PRISM Big Dye abus group sensu Nilsson (2000) (Agabus, Ilybius, Plat- Terminator Cycle Sequencing Ready Reaction Kit). ambus, and Ilybiosoma) which has a predominantly Sequencing reactions were purified by ethanol precipi- Holarctic distribution. No specimens of the two smaller tation and were electrophoresed on an ABI3700 se- genera within this group (, one species, and quencer. Sequence ambiguities were edited using the Hydronebrius, four species) could be obtained for study. Sequencher 3.0 software package (Gene Codes Due to the taxonomic homogeneity of the Ilybio- Corporation). soma, and the difficulty in separating the currently rec- ognized species (Larson et al., 2000), repeated specimens 2.2. Phylogenetic analysis of the only widespread species (I. seriatum) were in- cluded. Voucher specimens are kept in the collections of Sequences for COI were identical in length, and the Natural History Museum (London). there were no major differences in the length of 16S I. Ribera et al. / Molecular Phylogenetics and Evolution 30 (2004) 545–562 547

Table 1 Taxonomic coverage of the genera, subgenera, and species groups of Agabinae Genusa Subgenus Sp. group No. sp. No. sp. sampledb Agabus Acatodes arcticus 63 confinis 33 8 fuscipennis 5 — japonicus 12 3 lutosus 32 obsoletus 5 1(1) raffrayi 5 —

Agabus aeruginosus 3 — antennatus 11 clavicornis 31 disintegratus 31 falli 1 — labiatus 72 lineatus 21 punctulatus 5 — uliginosus 5 —

Gaurodytes adpressus 2 — affinis 94 ambiguus 5 4(2) aubei 11 brunneus 44 guttatus 23 11 nebulosus 62 paludosus 21 ragazzii 5 — striolatus 1 — taiwanensis 1 — tristis 94

Agametrus 7 — Andonectes 1 — Hydronebrius 4 — Hydrotrupes 1 —

Ilybiosoma seriatum 14 4(6) discicollis 2 —

Ilybius chalconatus 19 8 erichsoni 32 opacus 13 6(2) subaeneus 32 21

Leuronectes 5 —

Platambus americanus 2 — confusus 1 — glabrellus 22 maculatus 17 3 optatus 14 3 sawadai 5 — semenovi 4 — semivittatus 62 spinipes 4 —

Platynectes Australonectes 1 — Gueorguievtes 26 2 Platynectes 11 —

Total 361 107(11) a Classification and number of described species follows Nilsson (2001). b In brackets, multiple specimens of the same species. 548 I. Ribera et al. / Molecular Phylogenetics and Evolution 30 (2004) 545–562 sequences (see below). Alignment was performed in a for the combined data, were selected using Modeltest two step procedure (Phillips et al., 2000), aligning se- 3.06 (Posada and Crandall, 1998). To estimate relative quences in ClustalW (Higgins et al., 1992) followed by node ages we fitted ML branch lengths assuming a tree searches on the aligned matrix using parsimony molecular clock and compared the likelihood to that analysis in PAUP4.0b6 (Swofford, 1999). Three differ- obtained assuming no clock (Felsenstein, 1981). As the ent parameter settings were used in ClustalW: Align- ML ratio was significant (see Section 3), an ultrametric ment 1 (default values, i.e., gap cost 15, extension cost tree was estimated using the non-parametric rate 6.66), Alignment 2 (gap cost 2, extension cost 1), and smoothing (NPRS) method of Sanderson (1997), Alignment 3 (same parameters as 1, with some obvious as implemented in TreeEdit v1.0 (Rambaut, A. & mismatches manually corrected). PAUP searches con- Charleston, M.; evolve.zoo.ox.ac.uk). Approximate sisted of TBR heuristic searches of 10,000 replicates, calibration of absolute time was based on the standard with gaps coded as a fifth character (Giribet and molecular clock estimate of 2% divergence per Wheeler, 1999). Among these parameter combinations, million years (Myr) for mtDNA (Brower, 1994), cor- the preferred settings were selected according to the responding to a base rate (per branch) of 0.01 substi- maximum congruence among genes, as measured with tutions/site/Myr. the incongruence-length difference test, ILD (Farris Geographical data on the distribution of species were et al., 1994), and the modified ILD (WILD, Dow- obtained from Nilsson (2000, 2001) and Larson et al. ton and Austin, 2002; Wheeler and Hayashi, 1998; (2000). The number of transitions between major bio- Table 2). For the preferred alignment, analyses were geographical areas was reconstructed on the preferred repeated with gaps coded as missing characters, and tree using MacClade 4.0 (Maddison and Maddison, after the deletion of the specimens with missing data. 2000), coding ‘‘Holarctic’’ distributions as uncertain. For all the alignments, characters were reweighted a posteriori according to the rescaled consistency index (RC) (Farris, 1969) and a complete heuristic search 3. Results conducted on the best trees. Constraint trees for determining partitioned Bremer 3.1. Sequence data and trees obtained under different Support values (Baker and DeSalle, 1997) were gen- alignment parameters erated with Treerot (Sorenson, 1996), on both the equal weight and the reweighted trees obtained with Pairwise sequence divergences among any two taxa the preferred alignment. The significance of incon- ranged from zero for the 16S sequences of some species gruence among genes, and among the three codon to 29% for the COI gene, between Copelatus haemor- positions of the COI gene, was tested with the Parti- rhoidalis and Platambus sculpturellus. Maximum diver- tion Homogeneity Test (Swofford, 1999) as imple- gences within the ingroup were similar to those with the mented in PAUP. outgroups in both genes (17 and 18%, respectively, for the combined data). COI was not length variable, and 2.3. Rate variation and biogeographical analysis sequence length for 16S ranged from 508 (P. decem- punctatus, P. sculpturellus, I. pandurus, I. fraterculus, To estimate branch lengths, sequence variation of and I. subaeneus) to 513 nucleotides (I. montanus, the ingroup only was fitted by maximum likelihood I. bedeli, and I. hozgargantae). (ML) on the topology of the preferred parsimony tree. Accepting that congruence among genes can be Optimum ML models for the genes COI and 16S, and used as the criterion for the choice of the preferred

Table 2 Tree statistics and length of the aligned sequences Alig.a No. trees Cost ILD WILD CI RI No. Cha. Inf. Cha. 16S COI Com. 1 112 1601 4910 6672 161 0.024 0.16 0.56 1280 530 2 144 3146 4910 8317 261 0.031 0.17 0.47 1460 654 3 >5000 1570 4910 6620 140 0.021 0.16 0.56 1284 531 3m 4434 6517 0.16 0.55 1284 522 3r 288 6711 0.16 0.55 1284 531 Alig., Alignment; Com., combined; ILD, incongruence-length difference test (Farris et al., 1994; ILD ¼ length combined ) sum length individual sets); WILD, WheelerÕs ILD (Wheeler and Hayashi, 1998; WILD ¼ ILD/length combined); CI, consistency index; RI, retention index; No. Cha., length of the aligned sequence; Inf. Cha., number of informative characters. a 1, 2, 3: alignments, 1, gap cost 15, extension cost 6.66; 2, gap cost 2, extension cost 1; 3, gap cost 15, extension cost 6.66, modified by hand. m, Gaps as a missing character; r, species with missing data excluded (A. confinis, I. apicalis, I. bedeli, P. glabrellus, and A. nannup, see Section 2). I. Ribera et al. / Molecular Phylogenetics and Evolution 30 (2004) 545–562 549 parameter set (Wheeler, 1995), the optimal tree was dominated by the COI third positions. On the con- obtained from Alignment 3 (gap cost 15, extension trary, on the reweighted tree all pairwise correlations cost 6.66, manually modified), with the lowest ILD among PBS (three COI positions and 16S) are signif- and WILD (Table 2). As expected, character re- icantly positive, indicating a higher level of congru- weighting increased the congruence among the different ence. The support of the signal is also shifted towards partitions (16S and the three codon positions of COI), 16S and the first and second positions of COI (Table as measured by partitioned Bremer support (PBS) 4) with all partitions significantly and positively cor- (Table 4). In the equal weight tree, the correlation related with the total Bremer support value. This between the PBS of the COI third positions and that higher congruence among data partitions is reflected in of the second and with the 16S gene are significantly the lack of incongruence in the Partition Homogeneity negative (p < 0:05, 2 tails) (the average value of the Test (not significant, p 0:05). COI second positions is negative, Table 4). The PBS In the trees obtained using the preferred Alignment 3, values of the 16S gene are also negatively correlated and after reweighting the characters a posteriori (Figs. 1 with that of the second and first positions of COI, and 2; Table 3), Dytiscidae is recovered as monophy- indicating that the phylogenetic signal of the data is letic, with Agabinae (excluding Platynectes) sister to all

Table 3 Comparison of selected nodes in the phylogenetic analyses of Agabinae No.a Node 1 1R 2 2R 3 3R 3m 3mR 3r 3rR 1 Dytiscidae 1 0 1 0 0 1 1010 Agabinae + Colymbetinae 0 1 1 1 1 0 0001 2 Colymbetinae + Dytiscinae 0 0 0 0 0 1 0100 3 Agabinae excl. Platynectes 110101 0101 4 Ilybiosoma sister to Agabus s.lat. 1 1 0 0 0 1 0001 Ilybiosoma sister to Ilybius 000000 0100 Agabus s.lat. sister to Ilybius 0001f 1f 0 1f 010 5 Ilybiosoma 111111 1111 6 Ilybius 110111 1111 7 I. opacus group 1 1 1 1 1 1 1111 8 I. chalconatus–ericsoni group 1 1 1 1 1 1 1111 9 I. subaeneus s.str.c 110111 1111 10 Agabus s.lat.d 110101 1111 11 Gaurodytesb 011101 1f 111 12 A. guttatus group 1 1 1 1 1 1 1111 13 Acatodes + A. ambiguus group 1 1 1 1 1 1 1111 A.brunneus + A. affinise groups 0 0 0 1 1 0 1110 14 A.brunneus gr. sister to Acatodes 000001 0001 15 A. brunneus group 0 0 1 1 1 1 1111 Platambus 000000 0000 3 P.maculatus sister to Agabinae 00001f 1 1f 01f 0 P.maculatus gr. sister to Ilybius 110000 0100 1, 2, 3: alignments (1, gap cost 15, extension cost 6.66; 2, gap cost 2, extension cost 1; 3, gap cost 15, extension cost 6.66, modified by hand); R, reweighted according to the RC; m, gaps as a missing character; r, species with missing data excluded (A. confinis, I. apicalis, I. bedeli, P. glabrellus, and A. nannup, see Section 2). a Node numbers refer to Fig. 2. (In bold, preferred tree). b Excluding the A. brunneus and A. ambiguus groups. c Excluding I. apicalis and the I. angustior complex. d Including Platambus except the P. maculatus group. e Including A. lineatus. f With the inclusion of some additional taxa.

Table 4 Average partitioned Bremer support values of the shortest reweighted trees obtained with Alignment 3 (gap cost 15, extension cost 6.66, modified by hand) COI 1st pos. COI 2nd pos. COI 3rd pos. 16S rRNA Total Equal weight 5.44 6.63 )1.59 4.25 0.47 1.71 3.91 8.24 8.24 8.39 Reweighted 0.35 0.44 0.02 0.24 0.01 0.15 0.38 1.18 0.75 1.42 Cell contents: average SD. 550 I. Ribera et al. / Molecular Phylogenetics and Evolution 30 (2004) 545–562 other included species of Dytiscidae, and Colymbetinae among non-Agabinae Dytiscidae. Within Agabinae, the and Dytiscinae as single clades. Copelatinae is not re- Agabus group of genera sensu Nilsson (2000) (i.e., Ag- covered as monophyletic, nor Agabinae, as the two abus, Platambus, Ilybius, and Ilybiosoma) is, however, specimens of the genus Platynectes are always included recovered as monophyletic, although with low bootstrap

Fig. 1. Strict consensus of the 16 equally parsimonious trees obtained with the Alignment 3 (gap cost 15, extension cost 6.66, manually modified), after reweighting the characters according to the RC. Above branches, bootstrap values (only >50% shown); below branches, combined Bremer support values. *16S rRNA sequence identical to that of other species (see Appendix A). Nodes with 0.0 have positive Bremer support values lower than 0.05. I. Ribera et al. / Molecular Phylogenetics and Evolution 30 (2004) 545–562 551

Fig. 2. Phylogram of one of the most parsimonious optimal trees obtained with the Alignment 3 (gap cost 15, extension cost 6.66, manually modified), after reweighting the characters according to the RC. Branch lengths refer to Maximum Parsimony with equally weighted characters (note the long branches of the P. sculpturellus clade). Names of the groups refer to Nilsson (2001), with modifications (see text). Numbers inside nodes refer to numbers in Table 3. support. In alternative alignments Agabinae (excluded Closer inspection of the Agabus group reveals three Platynectes) is sister to Colymbetinae, more in agree- major clades, corresponding to the genera Ilybius, ment with traditional (Table 3). Ilybiosoma, and Agabus sensu lato (s.lat.). The latter 552 I. Ribera et al. / Molecular Phylogenetics and Evolution 30 (2004) 545–562 includes the genus Platambus, except for the P. macul- as a missing character, was highly significant atus group, which is either sister to all other members of (p 0:001). The exclusion of P. sculpturellus, P. gla- the Agabus group of genera or to Ilybius (Table 3). brellus, and P. princeps, with very long branches and in These three clades are very robust to alignment varia- an unlikely position in the tree (Fig. 2, see above), tions, with two of them (Ilybius and Ilybiosoma) recov- reduced the difference considerably, although it was ered in all trees, and with high Bootstrap and Bremer still highly significant. An ultrametric tree was thus support (Table 3, Fig. 1). The relationships among them estimated using the non-parametric rate smoothing are, however, less well defined. In the preferred tree (NPRS) method of Sanderson (1997) (see Section 2, Ilybiosoma is sister to Agabus s.lat. (Fig. 1), although all Fig. 3), excluding the three species with long branches. possible combinations of the three taxa were found For comparative purposes an ultrametric tree was also (Table 3). estimated by directly enforcing a molecular clock in The relationships within Ilybius and Ilybiosoma are PAUP. robust, and very similar in all trees obtained (Table 3, The best ML models estimated for the COI and 16S Fig. 1). Within Ilybius, the species of the I. opacus group genes separately were also complex GRT + I + G plus I. gagates are sister to the rest, which is divided in models, with Gamma shape parameters of 0.40 and the I. chalconatus plus erichsoni groups, and the I. su- 0.60, respectively. In all cases the ML ratio between baeneus group (with the exception of I. apicalis, which is the values obtained enforcing and not enforcing a sister to the species of the I. chalconatus–erichsoni molecular clock were highly significant (p 0:001). group) (Table 3, Fig. 1). Average length of the COI branches (as estimated with Within Agabus s.lat., three small clades form a basal the optimal ML model for COI on the most parsi- paraphyletic series, consisting of Platambus (excluding monious combined tree) was 0.071 substitutions/site, the P. maculatus group) and the subgenus Agabus s.str. while that of the gene 16S (estimated with the optimal (Figs. 1 and 2). The remaining species of Agabus are ML model for 16S on the most parsimonious com- divided in two large clades: the subgenus Gaurodytes bined tree) was only 0.015 substitutions/site. These (with the exclusion of the A. brunneus and A. ambiguus differences are reflected in the ultrametric trees esti- groups) and the subgenus Acatodes (plus the A. brunneus mated with NPRS for both genes separately. On and A. ambiguus groups). The species of the P. average, the branches of the COI gene were approxi- sculpturellus group plus P. princeps are included within mately twice as long after the NPRS correction (0.071 the species of the A. affinis group, but with no support vs. 0.125), compared to a factor of four (0.015 vs. and in a very long branch (Figs. 1 and 2), suggesting a 0.055) in 16S. The increase in branch length for the possible artifact of analysis. combined data was similar to the value for COI The position of the A. brunneus group is very (0.045–0.090). The total divergence for the ingroup (for variable, being in some trees sister to the A. affinis the basal node in Fig. 3, and excluding the P. group (Table 3). Within Gaurodytes, the A. guttatus sculpturellus clade) for the gene 16S alone in the NPRS group is always recovered and with strong support, in tree was 0.33 substitutions/site. Divergence among the most cases sister to A. aubei. The other species groups three main lineages (Ilybius, Ilybiosoma, and Agabus (A. tristis group and A. nebulosus group) are also re- s.lat.) was approximately 0.30 substitutions/site, and covered as monophyletic in all analyses. On the con- maximum divergence within each of these main lin- trary, within Acatodes (plus A. ambiguus group) the eages approximately 0.25 substitutions/site. For the resolution is minimal in all cases, being the only part COI gene these values were approximately doubled of the preferred tree with unresolved polytomies (0.70, 0.65, and 0.60, respectively). (Fig. 1). Maximum divergences for the combined data in the ultrametric tree of the ingroup estimated with NPRS 3.2. Rate of variation within and between clades and gene were ca. 0.60 substitutions/site (at the node splitting the regions species of the P. maculatus group from all others, de- termining the estimated age of the whole lineage), while The best ML model (as estimated with Modeltest, enforcing the clock directly without rate smoothing using the Hierarchical Likelihood Ratio Tests criteria) gave a maximum divergence of 0.35 substitutions/site. for the combined 16S and COI sequence for the in- Within the remaining taxa, the level of divergence was group only (i.e., Agabinae excluding Platynectes) was a very similar in the two main monophyletic lineages complex GTR + I + G, with estimated base frequencies, (Ilybius and Agabus s.lat.), although with different ab- among-site rate variation and a Gamma distribution solute estimates according to the method: 0.41 (Ilybius) shape parameter of 0.46. The ratio of the likelihood and 0.48 (Agabus s.lat.) substitutions/site for the values obtained enforcing and not enforcing a molec- NPRS, and 0.28 substitutions/site (both Ilybius and ular clock on one of the most parsimonious reweighted Agabus s.lat.) for the direct estimation of ML branch trees obtained with Alignment 3, and considering gaps length. I. Ribera et al. / Molecular Phylogenetics and Evolution 30 (2004) 545–562 553

Fig. 3. Ultrametric tree obtained with the non-parametric rate smoothing method of Sanderson (1997) using the topology of one of the optimal trees of Fig. 1 (ingroup only, excluded P. sculpturellus, P. glabrellus, and P. princeps), with initial branch lengths estimated through ML (see Section 2). Distribution of the species, E, East; W, West; H, Holarctic. Holarctic species were coded as uncertain (see Section 2). The scale was calibrated using a standard rate of ca. 0.01 substitutions/site/Myr per branch (Brower, 1994).

Even accounting for a slower rate of the gene 16S substitutions/site/Myr of Brower, 1994), there is an (as in, e.g., Gomez-Zurita et al., 2000, with 0.0038 inconsistency between the estimates of the age of di- substitutions/site/Myr instead of the standard 0.01 vergence of the gene 16S and that of the combined 554 I. Ribera et al. / Molecular Phylogenetics and Evolution 30 (2004) 545–562 data. Using this calibration in the gene 16S alone, age that these nodes lack support. The main phylogenetic estimates are ca. 30% older than with the combined and biogeographic conclusions are, however, not af- data. fected, as relationships within each of the main lin- eages are in general well supported and with more 3.3. Biogeographic separation of palearctic and nearctic stable topologies. lineages Most nodes with robust support in the preferred tree were also present in the majority of trees ob- The distributions of the species (Nearctic, Pale- tained with other alignments, either using equally arctic or Holarctic) was mapped on the preferred weighted or reweighted data. Nodes which were not phylogenetic tree to assess historical vicariance and universally recovered had minimal support in most dispersal patterns. Two of the three deep clades of cases, but even alternative resolutions of those nodes the Agabinae, Ilybius and Agabus s.lat., exhibit a produced a limited range of tree topologies. Some of basal separation of exclusively Nearctic and Palearctic these poorly supported nodes could be attributed to clades (the non-Nearctic species of the third main data artifacts, including long branches, which may be lineage, Ilybiosoma, could not be obtained for study, the reason for the unexpected position of the Plat- see Section 4). Within the Nearctic clade of Ilybius at ambus sculpturellus group, the possible polyphyly of least two species secondarily expanded their ranges to the genus Platambus, and/or the exclusion of Platy- the north Palearctic (I. opacus and I. wasastjernae, nectes from Agabinae. Fig. 3). Within the Palearctic lineage of Ilybius, fur- ther splits separate a Western Palearctic clade (I. 4.2. Phylogeny of agabinae chalconatus–erichsoni group), and a clade including all the species from the Eastern Palearctic, plus some According to our results, Agabinae (with the possible Western Palearctic and (secondarily) Nearctic species exclusion of Platynectes) is a monophyletic lineage not (I. subaeneus group). Among the Palearctic clade sister to Colymbetinae, in agreement with previous there are at least three inferred transitions to the conclusions based on morphological (Alarie et al., 2002; Nearctic, and two range extensions to Holarctic dis- Miller, 2001) and molecular (Ribera et al., 2002a) data. tributions. In Agabus s.lat. there are two basal clades The exclusion of Platynectes from Agabinae agrees with with an old Nearctic–Palearctic disjunction: the spe- our previous results based on a different gene, nuclear cies of an expanded A. labiatus group (including 18S rRNA (Ribera et al., 2002a). Although no mor- A. antennatus and A. serricornis), and the species of phological character supports the exclusion of Platy- the P. optatus group. The remaining Agabus split in: nectes from Agabinae, the highly plesiomorphic (1) a largely Western Palearctic clade (Gaurodytes condition of most Agabinae precludes any definitive excluding A. brunneus and A. ambiguus groups), with conclusion. only one expansion to an Holarctic range (A. tristis) Regarding basal relationships within Agabinae, the and a single transition to the Nearctic (A. semi- redefinition of the genera proposed by Nilsson (2000) is punctatus) and (2) a clade with mixed distributions, largely confirmed. Ilybiosoma and the expanded Ilybius Western and Eastern Palearctic, and Nearctic (A. that now includes several species groups formerly in brunneus group and Acatodes plus A. ambiguus group, Agabus are recovered as monophyletic with very strong Fig. 3). support. The genus Agabus is also recovered in the preferred tree, although with low support and only with the inclusion of some species groups of Platambus.On 4. Discussion the contrary, the genus Platambus is never recovered as monophyletic. The species of the P. maculatus group 4.1. Sequence length variability (part of the genus ‘‘Platambus ’’ in the restricted sense previous to Nilsson, 2000) are of uncertain position, The small difference in sequence length in 16S either as sister to all remaining Agabinae, or sister to rRNA resulted in alternative alignments and major Ilybius. The other species of the genus are included differences in tree topology, particularly at basal nodes within Agabus s.lat., or sister to it. The placement of defining the relationships among the main lineages. the P. sculpturellus group (the former genus The use of a direct optimization method, such us the plus P. princeps) remains uncertain, as their position in one implemented in the POY software (Gladstein and the preferred tree has no morphological support and is Wheeler, 1997; see Ribera et al., 2002a for an example clearly affected by long branches. Alarie (1995, 1998) in Dytiscidae) would perhaps provide a unique to- considered the species of Agabinus to be sister to Ag- pology, but the method is more demanding on com- abus + Ilybius, although the American species of Plat- putation given the size of the matrix (142 taxa with ambus (sensu Nilsson, 2000) were included within ca. 1300 nucleotides) and would not overcome the fact Agabus. In Miller (2001) Agabinus is placed in an I. Ribera et al. / Molecular Phylogenetics and Evolution 30 (2004) 545–562 555 unresolved polytomy formed by species of Platambus in potentially key groups missing (e.g., the P. sawadai agreement with Nilsson (2000). and P. semenowi groups, included in the former Plat- In what refers to the relationships within the three ambus in a restricted sense, Nilsson, 2000). The genus main generic groupings of Agabinae, the species was redefined based on weak morphological characters groups of Ilybius defined by Nilsson (2000) are in (a combination of prosternal process with lateral bead general well supported. The I. opacus group (sensu broadly inflated posterior of procoxae and/or mesoc- Larson, 1996) is recovered as monophyletic, and sister oxa widely separated, Nilsson, 2000). Yet this type of to I. gagates. The relatively isolated position of I. ga- evidence is not inferior to that defining other genera of gates is supported by its deviating morphology (cf. Agabinae which do have strong support in our anal- Miller, 2001). Together with I. larsoni (not sampled, ysis (such as Ilybiosoma, only characterized by the but morphologically very close to I. gagates, Fery and anteroventral spiniferous punctures along entire length Nilsson, 1993), these are the only Nearctic species of of the metatibia). Our results are more in agreement the I. chalconatus group as defined by Fery and Nils- with the concept of Platambus prior to Nilsson (2000), son (1993). The relationships within the I. opacus and although the lack of key taxa does not allow firm the I. subaeneus groups agree with the phylogenies conclusions. proposed in Larson (1996) and Ribera et al. (2001a), The subgenus Gaurodytes in our restricted sense respectively, with only minor differences. Within the (i.e., without the species of the A. brunneus group, of I. chalconatus–erichsoni group, the species included in uncertain position, and those of the A. ambiguus the I. chalconatus subgroup by Fery and Nilsson group, included together with Acatodes, see below) has (1993) (with the exclusion of I. gagates) are largely strong support, and the relationships of the main lin- paraphyletic, although other relationships are com- eages within it are almost identical in all trees obtained patible with their suggestions. under different parameters. The inclusion of the species The genus Ilybiosoma is always recovered as of the P. sculpturellus clade within the A. affinis group monophyletic. Its position within Agabinae is however is most likely an artifact of their long branch, as uncertain. In the preferred tree it is placed sister to noted. Agabus s.lat., although with low support. This is in The relationships within the A. affinis group (ex- agreement with Miller (2001), based on morphological cluded the P. sculpturellus clade) are in perfect agree- characters. In other trees it is sister to Ilybius, or basal ment with Nilsson and Larson (1990). However, this to Agabus plus Ilybius. The study of the European group is found to be nested within Agabus, and not species (which according to Nilsson, sister to the whole genus, as suggested by the same 2000 has some similarities with the species of Ilybio- authors. soma), and the non-Nearctic species of the genus (from The close relationship between the A. paludosus, A. Iran and Ethiopia) could help to resolve its placement. nebulosus, and A. guttatus groups was already suggested The only previous hypothesis of the relationships by Nilsson (1992a), in agreement with our results. The among the species of Ilybiosoma is that of Larson inclusion of the A. tristis group in this clade contrasts (1997), based in a phenetic ordination using morpho- with the results of Larson (1989), who suggested that it metric measures. The species groups thus defined agree is closely related to the species of the A. ambiguus with our results, although only two of them were group. included. We found Agabus aubei sister to the species of the A. Within Agabus, the subgenera defined by Nilsson guttatus group, as suggested by Balke et al. (1997). The (2000) are not recovered as monophyletic in the mo- relationships among the species of the A. guttatus group lecular analysis, although the main lineages largely are in agreement with Ribera et al. (2001b). The species conform to them. The basal clades include all species of with a distribution centered in the West Mediterranean Agabus s.str., plus some species groups of Platambus are sister to the species from the East (A. dilatatus and (P. optatus and P. semivittatus groups) in a paraphyletic A. glacialis, not included in Ribera et al., 2001b). The series. The first lineage is A. disintegratus plus the Moroccan A. alexandrae is found to be sister to A. species of the P. semivittatus group, followed by the A. maderensis, from Madeira. labiatus group plus the species with clubbed antennae The species of the A. brunneus group are found to be (A. antennatus and A. serricornis), a relationship sug- sister to the species of the subgenus Acatodes plus the gested by Angus (1984). This group is basal to Agabus A. ambiguus group. This position is however uncertain, s.lat., confirming the provisional placement of Nilsson with low nodal support and not recovered under (2000). other alignment parameters. An alternative position for A possible explanation for the observed polyphyly this group of species is as sister to the A. affinis group, of the species of Platambus (as defined in Nilsson, and thus within the remaining species of Gaurodytes, 2000) could be the relatively poor sampling of this a more likely relationship from a morphological point genus with respect to Agabus and Ilybius, with some of view. 556 I. Ribera et al. / Molecular Phylogenetics and Evolution 30 (2004) 545–562

The inclusion of the A. ambiguus group within the lineages within Agabinae would have occurred in the clade of the species with a bifid apex of the aedeagus Late Eocene. This is compatible with the fossil record, (i.e., Acatodes sensu Nilsson, 2000) was already sug- as the oldest known fossils attributed to Agabinae gested by Sharp (1882), who placed most of the spe- date from the Oligocene (see Nilsson, 2001 for a cies of the A. ambiguus, A. arcticus, and A. confinis checklist of fossil taxa). The basal splits within the groups in a single group (his ‘‘group 10’’). This rela- two main clades, Ilybius and Agabus s.lat., into Ne- tionship was also noted by Leech (1964), based on the arctic and Palearctic lineages apparently occurred chaetotaxy of the male tarsomeres. These similarities concurrently (Late Eocene–Early Oligocene), strongly were considered to be homoplasies by Larson (1994) suggesting a vicariant separation produced by the and Nilsson (2000) on defining the subgenera of Ag- same geological event. The study of the non-Nearctic abus, who consider the absence of the subapical spine species of Ilybiosoma (one Palearctic and one Ethio- on the aedeagus as a plesiomorphic character, ex- pian) would allow to determine if the same general cluding the species of the A. ambiguus group from this split Nearctic–Palearctic is reproduced in the third lineage. main lineage of Agabinae, and at the same level of Within the clade [Acatodes + A. ambiguus group], the divergence. relationships are largely unresolved, although some of The main Atlantic route of faunal interchange the species groups defined by Nilsson (2000) are par- between and North America, the Thulean tially recovered. The species of the A. confinis group are Bridge, closed some 50 Myr ago, presumably before placed in an unresolved polytomy, and the species of the the radiation of the main lineages of Agabinae A. japonicus group are paraphyletic, but the species of (Smith et al., 1994), and thus it is not likely that its the A. arcticus and ambiguus groups are recovered as closure had any effect on the evolution of the group. monophyletic. According to Larson (1991) the species of However, there was another Atlantic connection the A. confinis group (with the exclusion of A. elongatus, present until the Late Eocene (ca. 39 Myr), within the included in a different group) do not share any syna- time window of the early splits within Ilybius and pomorphy, and the group is defined by exclusion of all Agabus s.lat. This bridge connected Scandinavia with other likely monophyletic groups, i.e., it is likely to be north Greenland and Eastern North America through paraphyletic. the Canadian Arctic Archipelago (Sanmartin et al., 2001). Because of its northern position it is consid- 4.3. Historical biogeography of the species of Agabinae ered to have been suitable only for cold adapted organisms. The break-up of this land bridge in the The different estimates of ages from the 16S gene Late Eocene could have provided the initial vicari- alone and the combined data are surprising, since in ance event, but only if the ancestral ranges of the a previous study with a subset of Agabus analyzed Palearctic clades of Agabus and Ilybius were in the here, both calibrations agreed (Ribera et al., 2001b). western Palearctic. Although our results are compat- In that case the evolution of the combined sequence ible with this assumption, the small number of did not significantly deviate from clock-like variation Eastern Palearctic species sampled does not allow any and NPRS was not applied. The use of NPRS in the firm conclusion. present case could be the reason for this discrepancy, Another possibility is an initial connection through as the correction approximately doubled the average the Eastern Palearctic by way of the early Beringian branch lengths in COI, but increased them by a Bridges, which remained open until the Eocene–Oligo- factor of four in 16S. The NPRS is thus likely to cene boundary (35 Myr). The break-up of this connec- introduce deformations in the estimation of an ultr- tion was accompanied by a drastic change of global ametric tree, which are more pronounced in 16S with climatic conditions, and widespread vicariance events its slower rate of evolution and greater rate hetero- (Sanmartin et al., 2001). Both the separation of the geneity. To alleviate this effect it seems preferable to Thulean and the Beringian bridges are compatible with use the combined data for estimating the age of the the estimated date of the first vicariant split within Ily- lineages. However, even when using the combined bius and Agabus s.lat, but only when the older estima- data, the divergence estimates from the direct en- tion using the NPRS is accepted, or if the rate of forcement of a clock in PAUP are ca. 50% lower mtDNA divergence is lower than the standard 1% per than those obtained with NPRS, and hence these age branch per Myr. calibrations can be no more than a very rough ap- From the Mid-Jurassic (180 Myr) to late Tertiary proximation. (Oligocene, 30 Myr) the West and East Palearctic According to the level of divergence estimated in were divided by the Turgai Strait, located east of the the combined data using the NPRS correction, and Ural Mountains (Sanmartin et al., 2001). This sepa- assuming a standard clock rate of ca. 1% divergence ration was not complete, as the two regions were per branch per Myr, the diversification of the main connected on both ends of the strait, more extensively I. Ribera et al. / Molecular Phylogenetics and Evolution 30 (2004) 545–562 557 in the north. However, an additional marine trans- events if the divergence times are estimated directly gression during the Eocene increased the isolation, with the ML enforced clock), indicating that the lack which could have generally reinforced vicariance. The of apparent recent intercontinental vicariants is not inferred timing of the deep diversification within the due to the short time frame in which species separa- Palearctic groups of Ilybius and Agabus s.lat. is tions are studied. compatible with this event. Western Palearctic clades One of the most interesting results from the bio- are well defined in both groups, although for the geographical point of view is the sister relationship of hypothetical Eastern Palearctic lineage multiple range P. obtusatus and P. stygius, and their estimated level expansions to Holarctic or Nearctic distributions of divergence. This is compatible with the hypothesis would have to be postulated. The ancestral condition suggested by Nilsson (1996) of a relict Tertiary dis- of the supposedly Eastern Palearctic lineage, corre- tribution, even with the much lower direct estimate sponding to the I. subaeneus group and defined as the using ML (middle Miocene as opposed to Early sister of the well supported Western Palearctic clade, Oligocene with NPRS). This type of disjunct East is uncertain. In Ribera et al. (2001a) we proposed Palearctic–East Nearctic distribution is a common that the ancestral condition is Holarctic based on pattern among several groups of plants and , the distribution patterns of several basal species, but and is supposed to be a relict distribution of biota it is also possible that secondary range expansions inhabiting temperate subtropical forests which ex- would have eliminated information on their historical tended throughout the Palearctic during the early ranges. Tertiary, but which later became extinct in most of This general pattern of a basal Nearctic–Palearctic their range (Wen, 1999; Wu, 1983; see Sanmartin vicariance, followed by a second vicariance within et al., 2001 for a review). each of the main Palearctic clades between the West The main biogeographical patterns obtained from and the East, is in agreement with the results of our preferred tree are robust, as they do not depend on more comprehensive biogeographical analyses of the the exact resolution of basal relationships of the three Holarctic fauna (Enghoff, 1995; Sanmartin et al., main lineages within Agabinae. Most of the nodes 2001). The distributions within the subareas are in within the main lineages are very robust, and the un- general strongly phylogenetically conserved and in certainties affect only species poor lineages (P. macul- agreement with the results of Enghoff (1995) and atus group, A. brunneus group, and P. sculpturellus Sanmartin et al. (2001), as well as those for other group). Most species missing from the current study groups of animals (e.g., rodents, Riddle, 1998; birds, can be placed with confidence into morphologically Bohning-Gaese€ et al., 1998; Zink et al., 2000). Only well supported groups. Their inclusion could increase within the clades including Eastern Palearctic species the number of inferred range expansions or transitions, there is a relatively high frequency of range expan- but is unlikely to change the arrangement of the main sions and transitions (I. subaeneus group, Agabus lineages, and their reconstructed ancestral states. Per- subgenus Acatodes). This suggests a strong asymme- haps more important is the placement of a few major try in the main dispersal routes within the Holarctic, groups not sampled, as the species of Ilybiosoma from with the East Paleartic being the main source of outside the Nearctic, the exclusively Ethiopian A. rag- colonizers and the Nearctic and the West Palearctic azzii and A. raffrayi species groups, the Palearctic with lineages mostly restricted to their territories, P. sawadai and P. semenowi groups, and the genera also in agreement with previous biogeographical Hydrotrupes and Hydronebrius (Nilsson, 1992b, 2001). work (Enghoff, 1995; Pielou, 1979; Sanmartin et al., Their phylogenetic position based on morphology is 2001). uncertain (Beutel, 1994), but their placement could Most of the observed transitions between Palearctic potentially affect the inferred basal relationships of and Nearctic affect isolated species, and are compati- Agabinae. ble with a scenario of several subsequent connections through the Bering bridges, until the most recent (A. semipunctatus) at the end of the Miocene. The Acknowledgments Nearctic clade of most recent origin (late Tertiary) is within the subgenus Acatodes, which includes also We thank all the individuals listed in Appendix A most of the Eastern Palearctic species. We do not for providing material for study, and to Johannes infer any continental transitions during the Pliocene, Bergsten, Rob DeSalle and two anonymous referees except for a few apparent range expansions from for comments on an earlier version of the manuscript Palearctic or Nearctic to Holarctic in boreal species. and for editorial help. This project has been funded There is however frequent within-continent speciation; by a NERC grant (NER/B/S/1999/00003) to A.P.V. even based on our incomplete sampling we infer 22 and a Leverhulme Trust Special Research Fellowship speciation events in the last 5.2 Myr (more than 40 to I.R. 558 I. Ribera et al. / Molecular Phylogenetics and Evolution 30 (2004) 545–562

Appendix A. Studied material and sequence accession numbers

No. Speciesa Sp. group Dist. Country Collector 16S rRNA COI Ingroup AGABINAE Agabus Acatodes 1 A. anthracinus arcticus NUS I. Ribera and AY138602 AY138693 (Washington) A. Cieslak 2 A. arcticus arcticus H Sweden A.N. Nilsson AY138603 AY138694 3 A. sturmii arcticus P Sweden A.N. Nilsson AY138633 AY138723 4 A. confinis confinis H Sweden A.N. Nilsson AY138608 – 5 A. congener (8)b confinis P Moravia D. Boukal AY138609 AY138699 6 A. elongatus confinis H Sweden A.N. Nilsson AY138612 AY138702 7 A. kootenai confinis WN Canada I. Ribera and AY138594 AY138685 (Alberta) A. Cieslak 8 A. lapponicus (5) confinis P Sweden A.N. Nilsson AY138619 AY138709 9 A. matsumotoi confinis EP Japan J. Bergsten AY138622 AY138712 (Hokkaido) 10 A. setulosus confinis WP Sweden A.N. Nilsson AY138630 AY138720 11 A. thomsoni confinis H Canada I. Ribera and AY138595 AY138686 (BC) A. Cieslak 12 A. amoenus japonicus EP A.N. Nilsson AY138600 AY138691 (Astrakhan) 13 A. hummeli japonicus EP China J. Bergsten AY138616 AY138706 (Yunnan) 14 A. japonicus japonicus EP Kuriles Is. N. Minakawa AY138617 AY138707 15 A. lutosus lutosus WN US A. Cognato AY138620 AY138710 (California) 16 A. lutosus/griseipennis lutosus WN Canada I. Ribera and AY138593 AY138684 (BC) A. Cieslak 17 A. morosus (18) obsoletus WN US I. Ribera and AY138591 AY138682 (California) A. Cieslak 18 A. morosus2 (17) obsoletus WN US I. Ribera and AY138592 AY138683 (California) A. Cieslak Agabus (Agabus) 19 A. antennatus antennatus N Canada I. Ribera and AY138601 AY138692 (Alberta) A. Cieslak 20 A. serricornis clavicornis P Sweden A.N. Nilsson AY138629 AY138719 21 A. disintegratus disintegratus NUS W. Shephard AY071769 AY071795 (California) 22 A. labiatus labiatus WP UK I. Ribera AY138618 AY138708 23 A. bifarius labiatus H Canada I. Ribera and AY138605 AY138696 (Alberta) A. Cieslak 24 A. lineatus lineatus EP Russia A.N. Nilsson AY138611 AY138701 (Volgograd) Agabus (Gaurodytes) 25 A. affinis (26) affinis P Sweden A.N. Nilsson AY138598 AY138689 26 A. biguttulus (25) affinis P Sweden A.N. Nilsson AY138606 AY138697 27 A. semipunctatus affinis N Canada Y. Alarie AY138628 AY138718 (Ontario) 28 A. unguicularis affinis P UK I. Ribera AY138635 AY138725 29 A. ambiguus ambiguus N Canada Y. Alarie AY138599 AY138690 (Ontario) 30 A. ambiguus3 ambiguus N Canada I. Ribera and AY138596 AY138687 (BC) A. Cieslak 31 A. austinii ambiguus WN Canada I. Ribera and AY138604 AY138695 (BC) A. Cieslak 32 A. erythropterus ambiguus EN US K.B. Miller AY138613 AY138703 (New York) 33 Agabus sp. ambiguus N? US I. Ribera and AY138597 AY138688 (Washington) A. Cieslak 34 A. strigulosus ambiguus WN Canada I. Ribera and AY138632 AY138722 (BC) A. Cieslak 35 A. aubei aubei WP Corsica I. Ribera and AY039265 AY039277 A. Cieslak I. Ribera et al. / Molecular Phylogenetics and Evolution 30 (2004) 545–562 559

Appendix A (continued) No. Speciesa Sp. group Dist. Country Collector 16S rRNA COI 36 A. brunneus brunneus WP Spain A. Millan AY138607 AY138698 37 A. didymus brunneus WP Spain I. Ribera AF309274 AF309331 38 A. ramblae (39) brunneus WP Morocco I. Ribera AY138626 AY138716 39 A. rufulus (38) brunneus WP Corsica I. Ribera and AY138627 AY138717 A. Cieslak 40 A. alexandrae guttatus WP Morocco I. Ribera AY039257 AY039266 41 A. biguttatus guttatus P Spain I. Ribera AY039258 AY039267 42 A. biguttatus7 guttatus WP Canary Is. S. Ericsson AY039259 AY039270 (Gomera) 43 A. binotatus guttatus WP Corsica I. Ribera and AY039260 AY039271 A. Cieslak 44 A. dilatatus (46) guttatus P Cyprus K.W. Miller AY039261 AY039272 45 A. faldermanni guttatus P Iran D.T. Bilton AY138614 AY138704 46 A. glacialis (44) guttatus P Iran D.T. Bilton AY138615 AY138705 47 A. guttatus guttatus WP H. Fery AY039262 AY039273 48 A. heydeni guttatus WP Spain H. Fery AY039263 AY039274 49 A. maderensis guttatus WP Madeira D.T. Bilton AY138621 AY138711 50 A. nitidus guttatus WP Spain I. Ribera AY039264 AY039276 51 A. conspersus nebulosus P Spain I. Ribera AY138610 AY138700 52 A. nebulosus nebulosus P UK I. Ribera AY138624 AY138714 53 A. paludosus paludosus WP Spain I. Ribera AY138625 AY138715 54 A. bipustulatus tristis P Spain I. Ribera AF309275 AF309332 55 A. melanarius tristis WP UK J. Denton AY138623 AY138713 56 A. tristis tristis N,EP Canada I. Ribera and AY138634 AY138724 (BC) A. Cieslak 57 A. wollastoni tristis WP Madeira M. Drotz AY138636 AY138726 Ilybiosoma 58 I. lugens/perplexus seriatum WN US A. Cognato AF309271 AF309328 (California) 59 I. pandurus seriatum WN US I. Ribera and AY138648 AY138736 (California) A. Cieslak 60 I. perplexus seriatum WN US I. Ribera and AY138650 AY138738 (Washington) A. Cieslak 61 I. seriatum (62) seriatum NUS K.B. Miller AF309272 AF309329 (New York) 62 I. seriatum2 (61) seriatum N Canada I. Ribera and AY138655 AY138743 (BC) A. Cieslak 63 I. seriatum4 seriatum N Canada I. Ribera and AY138649 AY138737 (Alberta) A. Cieslak 64 I. seriatum5 (65) seriatum NUS I. Ribera and AY138652 AY138740 (California) A. Cieslak 65 I. seriatum6 (64) seriatum NUS I. Ribera and AY138651 AY138739 (California) A. Cieslak 66 I. seriatum7 seriatum N US (New York) C. Hernando AY138653 AY138741 67 I. seriatum8 (64) seriatum NUS Y. Alarie AY138654 AY138742 (California) Ilybius 68 I. albarracinensis chalconatus WP Portugal I. Ribera AF309277 AF309334 69 I. bedeli chalconatus WP Tunisia I. Ribera and AY138658 – A. Cieslak 70 I. chalconatus chalconatus WP Morocco P. Aguilera AF309278 AF309335 71 I. dettneri chalconatus WP Spain D.T. Bilton AY138659 AY138746 72 I. gagates chalconatus NUS C. Hernando AY138663 AY138750 (New Jersey) 73 I. hozgargantae chalconatus WP Spain I. Ribera AY138664 AY138751 74 I. montanus chalconatus WP UK I.Ribera AY138666 AY138753 75 I. satunini chalconatus WP Russia A.N. Nilsson AY138670 AY138757 (Volgograd) 76 I. erichsoni erichsoni H Sweden A.N. Nilsson AY138661 AY138748 77 I. subtilis erichsoni WP Sweden A.N. Nilsson AF309276 AF309333 78 I. euryomus opacus WN US I. Ribera and AY138662 AY138749 (California) A. Cieslak 79 I. hypomelas opacus WN Canada (BC) I. Ribera and AY138665 AY138752 A. Cieslak 560 I. Ribera et al. / Molecular Phylogenetics and Evolution 30 (2004) 545–562

Appendix A (continued) No. Speciesa Sp. group Dist. Country Collector 16S rRNA COI 80 I. opacus opacus H Sweden A.N. Nilsson AY138668 AY138755 81 I. vandykei opacus WN US I. Ribera and AY138671 AY138758 (California) A. Cieslak 82 I. walsinghami (83) opacus WN US I. Ribera and AY138656 AY138744 (California) A. Cieslak 83 I. walsinghami2 (82) opacus WN US I. Ribera and AY138660 AY138747 (California) A. Cieslak 84 I. walsinghami3 opacus WN US I. Ribera and AY138657 AY138745 (California) A. Cieslak 85 I. wasastjernae opacus H Sweden A.N. Nilsson AY138672 AY138759 86 I. aenescens subaeneus WP T. Berendok AF309294 AF309350 87 I. angustior subaeneus H Sweden A.N. Nilsson AF309289 AF309345 88 I. anjae subaeneus EP Kuriles Is. N. Minakawa AF309295 AF309351 89 I. apicalis subaeneus EP Kuriles Is. N. Minakawa AF309279 – 90 I. ater subaeneus WP UK I. Ribera AF309287 AF309343 91 I. biguttulus subaeneus EN US K.B. Miller AF309282 AF309338 (New York) 92 I. chishimanus subaeneus EP Kuriles Is. N. Minakawa AF309288 AF309344 93 I. crassus subaeneus WP Sweden A.N. Nilsson AF309297 AF309353 94 I. discedens subaeneus H Kuriles Is. N. Minakawa AF309296 AF309352 95 I. fenestratus subaeneus P Sweden A.N. Nilsson AF309291 AF309347 96 I. fraterculus subaeneus N Canada I. Ribera and AF309281 AF309337 (Alberta) A. Cieslak 97 I. fuliginosus subaeneus P UK I. Ribera AF309293 AF309349 98 I. guttiger subaeneus WP Finland T. Berendok AF309285 AF309341 99 I. meridionalis subaeneus WP Corsica I. Ribera and AF309292 AF309348 A. Cieslak 100 I. nakanei subaeneus EP Japan J. Bergsten AY138667 AY138754 (Hokkaido) 101 I. pleuriticus subaeneus NUS C. Hernando AY138669 AY138756 (Vermont) 102 I. quadriguttatus subaeneus WP UK I. Ribera AF309283 AF309339 103 I. quadrimaculatus subaeneus WN Canada I. Ribera and AF309290 AF309346 (BC) A. Cieslak 104 I. similis subaeneus WP Sweden J. Bergsten AF309284 AF309340 105 I. subaeneus subaeneus H Sweden A.N. Nilsson AF309286 AF309342 106 I. vittiger subaeneus WN/WP Sweden A.N. Nilsson AF309280 AF309336 Platambus 107 P. glabrellus glabrellus WN US I. Ribera and AY138637 – (California) A. Cieslak 108 P. sculpturellus glabrellus WN US I. Ribera and AY138638 AY138727 (California) A. Cieslak 109 P. lunulatus maculatus P Iran H. Fery and Elmi AY138674 AY138761 110 P. maculatus maculatus P Spain I. Ribera AF309273 AF309330 111 P. pictipennis maculatus EP Japan J. Bergsten AY138676 AY138763 112 P. obtusatus optatus EN US C. Hernando AY138675 AY138762 (Pennsylvania) 113 P. princeps optatus EP China J. Bergsten AY138677 AY138764 (Yunnan) 114 P. stygius optatus EP Japan J. Bergsten AY138679 AY138766 (Hokkaido) 115 P. semivittatus semivittatus NUS K.B. Miller AY138678 AY138765 (New York) 116 Platambus sp. N? US W.E. Steiner and AY138631 AY138721 (Maryland) J.M. Swearingen Platynectes (Gueorguievtes) 117 P. decempunctatus Australia C.H.S. Watts AY138680 AY138767 118 P. decempunctatus gr New Guinea C. Vaamonte AY138681 AY138768 Outgroup COLYMBETINAE 119 Colymbetes schildknechti Spain I. Ribera AF428197 AF428236 120 Meladema coriacea France P. Ponel AF428189 AF428209 121 Rhantus grapii UK I. Ribera AF428195 AF428234 I. Ribera et al. / Molecular Phylogenetics and Evolution 30 (2004) 545–562 561

Appendix A (continued) No. Speciesa Sp. group Dist. Country Collector 16S rRNA COI 122 R. suturalis Spain I. Ribera AF428196 AF428235 COPELATINAE 123 Aglymbus geotroi Oman Sattmann et al. AY138639 AY138728 124 Copelatus haemorrhoidalis UK I. Ribera AY071774 AY071800 125 C. utowaensis New Guinea M. Balke AY071775 AY071801 126 Lacconectus peguensis Myanmar H. Schillhammer AY071785 AY071811 COPTOTOMINAE 127 Coptotomus lenticus US K.B. Miller AY071776 AY071802 (New York) 128 C. loticus US C. Hernando AY138643 AY138731 (Georgia) DYTISCINAE 129 Graphoderus cinereus Spain I. Ribera AY138647 AY138735 130 dauricus Japan J. Bergsten AY138644 AY138732 (Hokkaido) 131 D. thianschanicus Russia A.N. Nilsson AY138645 AY138733 (Volgograd) 132 Eretes sticticus Iran H. Fery AY138646 AY138734 LANCETINAE 133 Lancetes lanceolatus Australia C.H.S. Watts AY138673 AY138760 134 L. varius Chile I. Ribera AY071784 AY071810 MATINAE 135 Allomatus nannup Australia L. Hendrich AY138640 – 136 Batrachomatus wingii Australia L. Hendrich AY138642 AY138730 137 B. daemeli Australia C.H.S. Watts AY138641 AY138729 AMPHIZOIDAE 138 Amphizoa insolens US A. Cognato AY071770 AY071796 (California) 139 A. lecontei US NHM AY071771 AY071797 (California) HYGROBIIDAE 140 Hygrobia australiasiae Australia C.H.S. Watts AY071779 AY071805 141 H. hermanni Spain I. Ribera AY071780 AY071806 142 H. maculata Australia D. Norton AY071781 AY071807 Dist., distribution (W, West; E, East; N, Nearctic; P, Palearctic; H, Holarctic). a Classification follows Nilsson (2001). b In brackets, species numbers with identical 16S rRNA sequences.

References Baker, R.H., DeSalle, R., 1997. Multiple sources of character information and the phylogeny of Hawaiian drosophilids. Syst. Alarie, Y., 1995. Primary setae and pores on the legs, the last Biol. 46, 654–673. abdominal segment, and the urogomphi of the larvae of Nearctic Beutel, R.G., 1994. On the systematic position of Hydrotrupes palpalis Colymbetinae (Coleoptera: : Dytiscidae) with an analysis Sharp (Coleoptera: Dytiscidae). Aquat. Insect. 16, 157–164. of their phylogenetic relationships. Can. Entomol. 127, Bohning-Gaese,€ K., Gonzalez-Guzm an, L.I., Bronw, J.H., 1998. 913–943. Constraints on dispersal and the evolution of the avifauna of the Alarie, Y., 1998. Phylogenetic relationships of Nearctic Colymbetinae Northern Hemisphere. Evol. Ecol. 12, 767–783. (Coleoptera: Adephaga: Dytiscidae) based on chaetotaxic and Brinck, P., 1948. Coleoptera of Tristan da Cunha. Results of the porotaxic analysis of head capsule and apendages of larvae. Can. Norwegian Scientific Expedition to Tristan da Cunha 1937–1938 Entomol. 130, 803–824. 17, 1–22. Alarie, Y., Archangelsky, M., Nilsson, A.N., 2002. Larval morphology Brower, A.V.Z., 1994. Rapid morphological radiation and conver- of genus Lancetes (Coleoptera: Adephaga: Dytiscidae): the hy- gence among races of the butterfly Heliconius erato inferred from pothesis of sister-group relationship with the subfamily Dytiscinae patterns of mitochondrial DNA evolution. Proc. Natl. Acad. Sci. revisited. Can. Entomol. 134, 467–501. USA 91, 6491–6495. Angus, R.B., 1984. A new Siberian Agabus Leach, with notes on the Dowton, M., Austin, A.D., 2002. Increased congruence does not other species with clubbed male antennae (Coleoptera: Dytiscidae). necessarily indicate increased phylogenetic accuracy—the behavior Entomol. Scand. 15, 193–202. of the Incongruence Length Difference Test in mixed-model Balke, M., in press. Dytiscidae. In: Beutel, R.G., Leschen, R.A.B analyses. Syst. Biol. 51, 19–31. (Eds.), Handbook of Zoology, vol. 3. De Gruyter, Berlin. Enghoff, H., 1995. Historical biogeography of the Holarctic: area Balke, M., Dettner, K., Hendrich, L., 1997. Agabus ‘‘Metronectes’’ relationships, ancestral areas, and dispersal of non-marine animals. aubei Perris: Habitat, morphological adaptations, systematics, Cladistics 11, 223–263. evolution, and notes on the phanaerofluicolous fauna (Coleoptera: Farris, J.S., 1969. A successive approximations approach to character Dytiscidae). Aquat. 19, 75–90. weighting. Syst. Zool. 18, 374–385. 562 I. Ribera et al. / Molecular Phylogenetics and Evolution 30 (2004) 545–562

Farris, J.S., Kallersj€ o,€ M., Kluge, A.G., Bult, C., 1994. Testing problem with a paraphyletic Agabus (Coleoptera: Dytiscidae). significance of incongruence. Cladistics 10, 315–320. Koleopt. Rndsch. 70, 17–36. Felsenstein, J., 1981. Evolutionary trees from DNA sequences: a Nilsson, A.N., 2001. World Catalogue of Insects, vol. 3, Dytiscidae. maximum likelihood approach. J. Mol. Evol. 17, 368–376. Apollo Books, Stenstrup. Fery, H., Nilsson, A.N., 1993. A revision of the Agabus chalconatus— Nilsson, A.N., Larson, D.J., 1990. A review of the Agabus affinis group and erichsoni—groups (Coleoptera: Dytiscidae), with a proposed (Coleoptera: Dytiscidae), with the description of a new species phylogeny. Entomol. Scand. 24, 79–108. from Siberia and a proposed phylogeny. Syst. Entomol. 15, Giribet, G., Wheeler, W.C., 1999. On Gaps. Mol. Phylogenet. Evol. 227–239. 13, 132–143. Phillips, A., Janies, D., Wheeler, W., 2000. Multiple sequence Gladstein, D.S., Wheeler, W.C., 1997. POY: the optimization of alignment in phylogenetic analysis. Mol. Phylogenet. Evol. 16, alignment characters. American Museum of Natural history, New 317–330. York. Pielou, E.C., 1979. Biogeography. John Wiley and Sons, New York. Gomez-Zurita, J., Juan, C., Petitpierre, E., 2000. The evolutionary Posada, D., Crandall, K.A., 1998. Modeltest: testing the model of history of the genus Timarcha (Coleoptera, Chrysomelidae) DNA substitution. Bioinformatics 14, 817–818. inferred from Mitochondrial COII gene and partial 16S rDNA Ribera, I., Barraclough, T.G., Vogler, A.P., 2001a. The effect of sequences. Mol. Phylogenet. Evol 14, 304–317. habitat type on speciation rates and range movements in aquatic Guignot, F., 1933. Les Hydrocanthares de France. Hygrobiidae, beetles: inferences from species-level phylogenies. Mol. Ecol. 10, Haliplidae, Dytiscidae et Gyrinidae de la France continentale, avec 721–735. notes sur les especes de la Corse et de lÕAfrique du Nord francßaise. Ribera, I., Hernando, C., Aguilera, P., 2001b. Agabus alexandrae sp.n. Miscellanea Entomologica, Toulouse. from Morocco, with a molecular phylogeny of the western Higgins, D.G., Bleasby, A.J., Fuchs, R., 1992. CLUSTAL V: Mediterranean species of the A. guttatus group (Coleoptera: Improved software for multiple sequence alignment. Comput. Dytiscidae). Insect Syst. Evol. 32, 253–262. Appl. Biosci. 8, 189–191. Ribera, I., Hogan, J.H., Vogler, A.P., 2002a. Phylogeny of Hydra- Larson, D.J., 1989. Revision of North American Agabus Leach dephagan water beetles Inferred from 18S rDNA sequences. Mol. (Coleoptera: Dytiscidae): introduction, key to species groups, and Phylogenet. Evol. 23, 43–62. classification of the ambiguus-, tristis-, and articus-groups. Can. Ribera, I., Beutel, R.G., Balke, M., Vogler, A.P., 2002b. Discovery of Entomol. 121, 861–919. Aspidytidae, a new family of aquatic Coleoptera. Proc. R. Soc. Larson, D.J., 1991. Revision of North American Agabus Leach Lond. B 269, 2351–2356. (Coleoptera: Dytiscidae): elongatus-, zettertedti-, and confinis- Ribera, I., Nilsson, A.N., 1995. Morphometric patterns among diving groups. Can. Entomol. 123, 1239–1317. beetles (Coleoptera: Noteridae, Hygrobiidae, Dytiscidae). Can. J. Larson, D.J., 1994. Revision of North American Agabus Leach Zool. 73, 2343–2360. (Coleoptera: Dytiscidae): lutosus-, obsoletus-, and fuscipennis- Riddle, B.R., 1998. The historical assembly of continental biotas: Late groups. Can. Entomol. 126, 135–181. Quaternary range-shifting, areas of endemism, and biogeographic Larson, D.J., 1996. Revision of North American Agabus Leach structure in the North American fauna. Ecography 21, 437–446. (Coleoptera: Dytiscidae): the opacus-group. Can. Entomol. 128, Sanderson, M.J., 1997. A nonparametric approach to estimating 613–665. divergence times in the absence of rate constancy. Mol. Biol. Evol. Larson, D.J., 1997. Revision of North American Agabus Leach 14, 1218–1231. (Coleoptera: Dytiscidae): the seriatus-group. Can. Entomol. 129, Sanmartin, I., Enghoff, E., Ronquist, F., 2001. Patterns of 105–149. dispersal, vicariance and diversification in the Holarctic. Biol. J. Larson, D.J., Alarie, Y., Roughley, R.E., 2000. Predaceous Diving Linn. Soc. 73, 345–390. Beetles (Coleoptera: Dytiscidae) of the Nearctic Region, with Sharp, D., 1882. On aquatic carnivorous Coleoptera or Dytiscidae. Emphasis on the Fauna of Canada and Alaska. NRC Research Scient. Trans. R. Dubl. Soc. 2, 179–1003. Press, Ottawa. Simon, C., Frati, F., Beckenbach, A.T., Crespi, B., Liu, H., Flook, P., Leech, H.B., 1964. A new species of Agabus from the Death Valley 1994. Evolution, weighting, and phylogenetic utility of mitochon- region of California (Coleoptera: Dytiscidae). Coleopt. Bull. 18, drial gene sequences and a compilation of conserved polymerase 79–82. chain reaction primers. Ann. Entomol. Soc. Am. 87, 651–701. Lund, D.H., Zhang, D.X., Szymura, J.M., Hewitt, G.M., 1996. The Smith, A.G., Smith, D.G., Funnell, B.M., 1994. Atlas of Mesozoic and insect cytochrome oxidase I gene: evolutionary patterns and Cenozoic Coast Lines. Cambridge University Press, Cambridge. conserved primers for phylogenetic studies. Insect. Mol. Biol. 5, Sorenson, M.D., 1996. TreeRot. University of Michigan, Ann Arbor. 153–165. Swofford, D.L., 1999. PAUP*: Phylogenetic Analysis using Parsi- Maddison, W.P., Maddison, D.R., 2000. MacClade, Version 4.0. mony. Version 4.0b2a. Sinauer Associates, Sunderland, MA. Sinauer, Sunderland, MA. Vogler, A.P., DeSalle, R., Assmann, T., Knisley, C.B., Schultz, T.D., Miller, K.B., 2001. On the phylogeny of the Dytiscidae (Insecta: 1993. Molecular population genetics of the endangered tiger , Coleoptera) with emphasis on the morphology of the female Cicindela dorsalis (Coleoptera: Cicindelidae). Ann. Entomol. Soc. reproductive system. Insect. Syst. Evol. 32, 45–92. Am. 86, 142–152. Nilsson, A.N., 1992a. Another synonym of Agabus faldermanni Wen, J., 1999. Evolution of Eastern Asian and Eastern North America Zaitzev, with a discussion of the guttatus-, nebulosus- and paludo- disjunct distributions in flowering plants. Annu. Rev. Ecol. Syst. sus-groups (Coleoptera: Dytiscidae). Koleopt. Rndsch. 62, 43– 30, 421–455. 45. Wheeler, W.C., 1995. Sequence alignment, parameter sensitivity, and Nilsson, A.N., 1992b. A revision of Afrotropical Agabus Leach the phylogenetic analysis of molecular data. Syst. Biol. 44, 321–331. (Coleoptera, Dytiscidae), and the evolution of tropicoalpine super Wheeler, W.C., Hayashi, C.Y., 1998. The phylogeny of the extant specialists. Syst. Entomol. 17, 155–179. chelicerate orders. Cladistics 14, 173–192. Nilsson, A.N., 1996. A redefinition and revision of the Agabus optatus- Wu, Z., 1983. On the significance of Pacific intercontinental discon- group (Coleoptera, Dytiscidae); an example of Pacific interconti- tinuity. Ann. Missouri Bot. Gard. 70, 577–590. nental disjunction. Entomol. Basil. 19, 621–651. Zink, R.M., Blackwell-Rago, R.C., Ronquist, F., 2000. The shifting Nilsson, A.N., 2000. A new view on the generic classification of the roles of dispersal and vicariance in biogeography. Proc. R. Soc. Agabus-group of genera of the Agabini, aimed at solving the Lond. B 267, 497–503.